The clicking of correct tiles or entering the right text is merely a small part of the test after all

In practice, there are several types of CAPTCHAs: text-based, image-based, audio-based and behavior-based.

Computer scientist writing an article about captchas doesn’t know that proof-of-work captchas exist… Such a joke.

Of course those don’t do much against automated fake form filling, but against DDoS they’re THE solution.

Proof of work (PoW) captchas make the (computing) cost for the attacker exponentially higher than for the website. Basically the website creates a challenge, which has to be solved by the client/attacker before getting access to content. Best of all the website can set the difficulty of the challenge to anywhere from instant to seconds to solve, so normally the users don’t even notice it (as it runs in the background) but once someone starts DDoS’ing the difficulty goes up.

That just illustrates a more fundamental problem

So instead of spending money on extravagant expenses they could have fired 5% less people this round… Compared to what the money is spent on that still sounds good.

Though it indeed makes little difference in the big picture, this is still a bad image.

Best of all captchas which require human interaction are completely redundant anyways. From a security perspective anyways (blocking bots generating huge amounts of traffic)… For training your next LLM however…

For me it’s that a game I regularly play really needs their rootkit to run before they allow me to start it… If that ever changes or I stop playing it I’ll take a long hard look at Linux.

Lock the settings app behind a passcode too, there’s ways to block things like settings… Not sure if that works for individual screens in there though

I mean, duh? With proof of work captchas existing, there’s no reason to have those image selection captchas… Ever…

How those work is by having the server generate a puzzle. Server side this is cheap to generate, while client side solving is “hard”. The server can even choose the difficulty of the puzzle, and even set it dynamically. This means that when your website is under light load the captcha can be really easy/fast to solve. If your website is under attack however the captcha can be set to take seconds to solve.

Train system is not exactly viable here compared to using a car (Belgium)

Edit: but yeah the rest is about right

Same here, sure there’s hacks and workarounds that don’t require root… But still why the extra step…

I just want my window animation speed to be faster, why does that disqualify me from reading stuff sent to my government mailbox.

Ikr, like I don’t need a full feature full stack framework… I just want my tech demo to not look like it was made in the 80s without spending hours. (I’m mostly a backend dev)

Small bits like caps can’t get sorted for recycling for some reason, so they’re just “waste” instead of recyclable

Plenty of paywalled sites have workarounds (Bypass Paywall Clean extension basically does that for you)

So you feel bad about slapping a mosquito? What about skipping a rock?

It is interesting to consider where we draw the line of what we define as an ‘intelligent animal’…

On top of the reason the top level comment gave (easy for law enforcement) it also allows for better data collection (linking your activity to your account, no matter where, how or when it is recorded)

Valorant kernel level driver managed to fuck up my whole system every few updates though. It’s managed to freeze my whole pc 2 times, cause a BSoD 3 times… And that’s only talking about last week.

Almost as if running unstable software designed to be intrusive to other programs on a kernel level was not a great idea. But hey less cheaters right??? Oh wait no, turns out it still doesn’t stop cheaters.

No way this wasn’t written by ChatGPT

That’s assuming the old comments are actually overwritten instead of just marked as ‘old’

New models are like really good at context, the amount of input that can be given to them has exploded (fairly) recently… So you can give whole datasets or books as context and ask questions about them.